1. Technical Field
The present invention relates to a blind via structure, and associated laser ablation methods of formation, that includes a blind via within a photoimageable dielectric (PID) layer on a substrate, such that the sidewall of the blind via makes an obtuse angle with the blind end of the blind via.
2. Related Art
Electronic structures commonly include blind vias coupled to a dielectric substrate. A blind via is characterized by an open end and a closed end (xe2x80x9cblind endxe2x80x9d). If the sidewall of a blind via is plated with metal and subsequently treated with solder, the blind via may serve to electrically couple a circuitization layer on the substrate with such electrical interconnects as a ball grid array (BGA) or a controlled collapse chip connection (C4) solder ball. The BGA may be coupled to an electronic carrier such as a circuit card. The C4 solder ball, which may have any utilizable geometrical shape such as a spherical or columnar structure, may be coupled to an electronic assembly such as a chip. A blind via may be formed within a dielectric substrate by mechanical drilling or by laser ablation. With either method of blind via formation, each blind via must be individually and separately formed. This is costly and time-consuming for electronic structures that require numerous blind vias on the same dielectric substrate.
A known method of forming blind vias simultaneously includes use of a photoimageable dielectric (PID) material. If a PID layer is formed on a substrate, blind vias may be formed within the PID layer by a known photoimaging process that cross links (i.e., polymerizes) photoexposed PID material and allows multiple blind vias within the PID layer to be formed simultaneously. This known process is illustrated in FIGS. 1-4.
FIG. 1 shows a front cross-sectional view of a PID layer 12 on a substrate 10 for the purpose of forming one or more blind vias within the depth of the PID layer 12. In particular, FIGS. 1-4 depict the formation of two blind vias of different widths. The PID layer 12 may be formed on the substrate 10 by any known method in the art. A preliminary step in the process of blind via formation is baking the PID layer 12 (on the substrate 10) at a temperature, such as between 85 and 95xc2x0 C., that will evaporatively remove residual solvents within the PID material. The residual solvents are impurities that may interfere with cross linking of the subsequently photoexposed PID material.
In FIG. 2, a mask layer 14 has been placed on the PID layer 12. The mask layer 14 includes an opaque region 16 and an opaque region 18. The opaque region 16 and the opaque region 18, which may be formed inter alia by emulsions on a glass-plate representation of the mask layer 14, are located above portions in the PID layer 12 where blind vias will be subsequently formed. The mask layer 14 also includes a transparent region 30. The PID layer 12 includes a blocked region 26 under the opaque region 16, a blocked region 28 under the opaque region 18, and an unblocked region 34 under the transparent region 30.
Referring to FIG. 3, blind vias will be subsequently formed from the blocked region 26, the first blocked region 28, and portions (called xe2x80x9csoftxe2x80x9d regions) of the unblocked region 34. A soft region includes a portion of the unblocked region 34 which contacts a blocked region and is geometrically positioned to be subsequently etched away, as will be explained infra. As shown in FIG. 3, the unblocked region 34 includes a soft region 27 which contacts the blocked region 26, a soft region 29 which contacts the blocked region 28, and a hard region 32 which include all portions of the unblocked region 34 which are not soft regions. As will be described infra and depicted in FIG. 4, a first blind via 46 will be subsequently formed from the blocked region 26 and the soft region 27, and a second blind via 48 will be subsequently formed from the blocked region 28 and the soft region 29.
FIG. 3 depicts an xe2x80x9cexposurexe2x80x9d step in which a light source 20 directs light 22 of a suitable wave length, such as ultraviolet radiation, onto the mask layer 14. The light 22 passes through the transparent region 30 of the mask layer 14 into the unblocked region 34 of the PID layer 12. The light 22 initiates a cross-linking process which polymerizes the PID-exposed material in the unblocked region 34. The rate and magnitude of cross linking decreases with increasing distance in the direction 60 as measured from the top surface 36 of the unblocked region 34. PID material that is not cross linked (i.e., PID material in the blocked region 26 and in the blocked region 28) tends to dissolve, or be washed away, if contacted by a developer solution. PID material that is cross linked (i.e., PID material in the unblocked region 34, especially in the hard region 32) tends to resist being dissolved if contacted by a developer solution. After the preceding exposure step, the mask layer 14 is removed.
After the exposure step, the PID layer 12 may optionally be allowed to stand at room temperature for a period of time, typically between a half-hour and two hours. This optional xe2x80x9cstandingxe2x80x9d step relates to the fact that the cross linking process in the unblocked region 34 may continue to occur even after exposure of the mask layer 14 to the light 22 has been terminated. The time duration of the standing step, and whether the standing step should be utilized, depends on the cross-linking characteristics of the PID material in relation to the light 22; i.e., on the rate of crosslinking that occurs both during and after exposure of the mask layer 14 to the light 22 in consideration of the amount of cross linking that prevents light-exposed PID material from being washed away by a subsequently-applied developer solution.
After the mask layer 14 is removed, and after the exposure step (or after the standing step if elected), the PID layer 12 is exposed to a developer solution, such as by spraying, which washes away the PID material that has not been cross linked or sufficiently cross linked. In this xe2x80x9cdevelopingxe2x80x9d step, the washing away of the non-polymerized PID material in the blocked region 26, and of the insufficiently polymerized PID material in the soft region 27, results in formation of the first blind via 46 as shown in FIG. 4. In FIG. 3, the PID material in the soft region 27 is insufficiently cross linked to resist being washed away because of the attenuation of cross linking with increasing distance in the direction 60, as stated previously. In effect, the less cross-linked PID material at the bottom of the soft region 27 is more susceptible to being washed away by developer solution than is the more cross-linked PID material at the top of the soft region 27. Thus, after the developer solution washes away the non-polymerized PID material in the blocked region 26, the residual developer solution within the blocked region 26 contacts and washes away the insufficiently cross linked material in the soft region 27. Similarly, the washing away of the non-polymerized PID material in the blocked region 28, and of the insufficiently polymerized PID material in the soft region 29, results in formation of the second blind via 48 as shown in FIG. 4. A blind via formed as described above, such as the first blind via 46 or the second blind via 48, is called a xe2x80x9cphotoviaxe2x80x9d. After the preceding developing step, cleaning may be accomplished, if necessary, by any known method such as by rinsing the PID layer 12 with water.
FIG. 4 depicts a sidewall 56 of the first blind via 46 as making an angle xcex81 with respect to the blind end (i.e., bottom bounding surface) 52 of the first blind via 46. Similarly, the sidewall 58 of the second blind via 48 makes an angle xcex82 with respect to the blind end 54 of the second blind via 48. xcex81 and xcex82 are acute angles (i.e., less than 90 degrees) and result from the washing away of the soft region 27 and the soft region 29, respectively, as described supra in relation to FIG. 3. Definitionally, an acute-angled blind via is a blind via having an acute-angled sidewall such as the sidewall 56 of the first blind via 46, while an obtuse-angled blind via is a blind via having an obtuse-angled sidewall such as sidewall 76 of obtuse-angled blind via 49 with an associated obtuse angle 03 as shown in FIG. 6 (described infra) . Thus, the first blind via 46 and the second blind via 48 are acute-angled blind vias. Aside from an undercut region 57, at the bottom of the sidewall 56 in FIG. 4, the shape of the sidewall 56 of the first blind via 46 is typically linear, because the variation of the rate and magnitude of cross linking with distance in the direction 60 is continuous and approximately linear. Similarly, the shape of the sidewall 58 of the second blind via 48 is typically linear, aside from an undercut region 59 at the bottom of the sidewall 58. The undercut regions 57 and 59 may be formed if overdeveloping occurs. If underdeveloping occurs, however, there will be no undercut regions. With underdeveloping, excess PID material may remain at the bottom of the via sidewalls, resulting in a reduction in the width of the blind end 52 of the first blind via 46 and, similarly, a reduction in the width of the blind end 54 of the second blind via 48.
The magnitude of xcex81 depends on the rate of attenuation of cross linking as measured in the direction 60 from the top surface 36 of the unblocked region 34. Accordingly, xcex81 can be increased if the attenuation of cross linking with depth is diminished. Nonetheless, there is a limit as to how large xcex81 can become, because as a top portion of the unblocked region 34 becomes cross linked, the top portion becomes increasingly opaque to the light 22 that triggered the cross linking, which limits the amount of cross linking that can occur below the top portion. Similarly, there is a limit as to how large xcex82 can become. The actual upper limit to xcex81 and xcex82 depends on the cross linking characteristics of the PID material and on such geometric factors as a thickness (d) of the unblocked region 34.
A photovia, such as the first blind via 46 and the second blind via 48, typically undergoes subsequent processing, including metallic plating of conductive metal followed by application of solder to the metallic plating within the blind via. Known methods of plating typically include applying a seeding material on a blind via sidewall, followed by electroless plating of a thin layer (e.g., about 15-100 microinches) of metal such as copper. The seeding is necessary, because a metal such as copper does not directly plate on the organic material of a PID. A currently used seeding material is palladium. Seeding may alternatively be accomplished by sputtering a thin layer of copper (e.g., about 5000 xc3x85) on the blind via sidewall, wherein the sputtered layer of copper serves as a thin layer of seeding material for subsequent plating of a thicker layer of copper (e.g., 1 mil). After the thin layer of copper is formed as described supra, a relatively thicker layer (e.g., 1 mil) of copper is applied by electroplating. In addition to being formed on the first blind via 46 and the second blind via 48, thin and thick layers of copper may also be formed on the top surface 36 of the unblocked region 34, followed by creation of circuit features from the formed copper layers on the top surface 36. After the circuit features are created, there may be additional plating of the copper layers on the top surface 36 and/or the copper layers on the first blind via 46 and the second blind via 48. Such additional plating may include inter alia plating a layer of nickel over the thick layer of copper followed by plating a layer of gold over the layer of nickel. Following the additional plating, a solder bump may be formed in the blind via by inserting solder paste with solder flux in the blind via and reflowing the solder paste to form the solder bump.
The aforementioned method of forming photovias can be advantageously used to form many photovias simultaneously, such as 10,000 or 100,000 photovias. Nonetheless, the acute angles xcex81 and xcex82 present various problems relating to difficulty of accessing the bottom portion of the blind via sidewalls and the fact that the open end of the blind via is narrower than the blind end of the blind via. A first problem is that a blind via sidewall having an acute angle, such as xcex81 and xcex82 may be difficult to reliably plate, regardless of whether the seeding is with palladium or by sputtering of copper, because the portion of the blind via sidewall at the bottom of the sidewall is difficult to access by the seeding and plating material. As a result, metallic plating of the via thins out near the bottom of the blind via sidewall. Thin plating is susceptible to degradation during thermal cycling. The plating is particularly difficult and unreliable if aspect ratio of the blind via (i.e., axial depth/width of open end) is greater than about 1. In FIG. 4, the first blind via 46 has an aspect ratio xcex11 of d/w1, where w1 is the width of the open end of the first blind via 46 and d is the axial depth of the first blind via 46. The second blind via 48 has an aspect ratio xcex12 of d/w2, which is less than xcex11 where w2 is the width of the open end of the second blind via 48 and d is the axial depth of the second blind via 48. If xcex12 greater than 1 and xcex11 less than 1, then the second blind via 48 may be problematic, while the first blind via 46 is not likely to be problematic. In current practice, PID layers are typically about 2 to 3 mils thick with consequent photovia axial depths of about 2 to 3 mils. If the blind via depth (d) is about 3 mils in FIG. 4, then the width (w1) of the open end of the first blind via 46, and the width (w2) of the open end of the second blind via 48, should each be at least about 3 mils in order to assure reliable plating. Thus with a PID layer thickness of about 3 mils, the aforementioned method of photovia formation with an acute-angle sidewall is unreliable for formation of blind vias if the open end has a width of less than about 3 mils.
A second problem with the aforementioned method of blind via formation characterized by an acute-angled sidewall relates to formation of a solder bump within the blind via. Formation of a solder bump typically involves introduction of solder paste and solder flux into the blind via after formation of a metallic coating on the blind via sidewall. When the solder melts during reflow, the solder flux becomes volatile and forms gases which must escape from the blind via; otherwise trapped flux gases will pressurize within the blind via and push the liquid solder material out of the blind via. Unfortunately, an acute-angled sidewall with an adjoining undercut region in a blind via having an aspect ratio greater than about 1 tends to trap such gases, resulting in the aforementioned propensity to push solder out of the blind via.
A third problem relates to undercut regions that sometimes form at the bottom corners of acute-angled blind vias, such as undercut regions 57 and 59 in FIG. 4. An undercut region limits access to a blind via sidewall, such as access by a seeding solution or a plating solution. An undercut region also tends to trap chemical solutions, which result in less uniform plating within the undercut region. Additionally, an undercut region may trap hydrogen gas that is released by the electroless plating of copper . Thus, an undercut region add to reliability concerns relating to subsequent processing following the blind via formation.
A practical method is needed for forming an obtuse-angled blind via, as opposed to an acute-angled blind via, for situations where an acute-angled blind vias impairs reliability, such as where an aspect ratio of the blind via is greater than about 1.
The present invention provides an electrical structure, comprising: a substrate, a photoimageable dielectric (PID) layer on the substrate, and a blind via within the PID layer, said blind via having a obtuse-angled sidewall.
The present invention provides a method of forming an electrical structure, comprising: providing a substrate, forming photoimageable dielectric (PID) layer on the substrate, and forming a blind via having an obtuse-angled sidewall.
The present invention has the advantage of generating an obtuse-angled blind via in a PID layer.
The obtuse-angled blind via of the present invention has the advantage of eliminating undercut regions located where an acute-angled sidewall interfaces with a blind end of the acute-angled blind via.
The obtuse-angled blind via of the present invention has the advantage of mitigating reliability problems associated with plating of blind vias and formation of solder bumps within the blind vias. Such reliability problems relate to, inter alia, nonuniformity of seeding, nonuniformity of sputtering, nonuniformity of plating, diminished adhesion between a seed layer and a plating layer, trapping of gas during electroless plating of copper, trapping of gas during formation of solder bumps.
The present invention has the advantage of providing both acute-angled blind vias and obtuse-angled blind vias within a given PID layer, so as to combine the efficiency of simultaneously generating multiple acute-angled blind vias with the selectivity of generating only those obtuse-angled blind vias as are necessary to mitigate reliability problems in subsequent processing steps.
The present invention has the advantage of providing a plurality of stacked PID layers on a given surface of a substrate such that each of the stacked PID layers includes an obtuse-angled blind via.
The present invention has the advantage of providing a plurality of stacked PID layers on a given surface of a substrate such that an obtuse-angled blind via penetrates each of the stacked layers.
The present invention has the advantage of providing a first PID layer on a first surface of a substrate and a second PID layer on a second surface of the substrate, such that the first PID layer and the second PID layer each include an obtuse-angled blind via.